RCBldEng - a computationally efficient RC network modeling method and engine for multizone building simulation

As there is a lack of resistance-capacitance (RC) modeling method and framework for universal applicability and standardized implementation to multizone building simulation practices, a novel hybrid RC modeling approach that combines physics-based principles (forward modeling) with real-world data calibration (inverse modeling) for building energy simulation is presented in this study. A simulation engine called RCBldEng that combines theoretical building physics with real world calibration capability is developed. Four RC model configurations are proposed and evaluated for different building types and thermal behaviors (4R1C, 6R1C, 7R1C, and 7R2C) with increasing model complexity. The EnergyPlus simulations of two prototype buildings and an educational building case study in the real world are used to test and validate the performance of the simulation engine. The results indicate that the inclusion of interzonal thermal coupling and dual capacitance mechanisms (7R1C and 7R2C) substantially improves prediction accuracy leading to R2 values of up to 98.78% for office building cooling load prediction against EnergyPlus. For simplified building simulation model, configurations such as the 4R1C model are suitable, but the 7R2C model can better represent the thermal behavior of buildings with high effective thermal mass and multiple zones. Moreover, the differential evolution algorithm turns out to be an effective choice for model calibration for existing buildings, and real-world operational uncertainties due to seasonal variations can be important bias sources. It is shown that the computational efficiency of RCBldEng can be a competitive candidate in performing building simulation for preliminary design stage and optimization applications.